Fix chunking for .ts and .tsx files (#73)

Fixes for Gemini embeddings (#71)

Add an LLM retriever (#70)

Auto-format code with isort and black

add support for google gemini embeddings as an embedder (#56)

Auto-format code with isort and black

Option for multi-query retrieval (#51)

Auto-format code with isort and black

Better defaults for remote.yaml

Default --retrieval-alpha to 1.0

Fix typo in system prompt (#52)

add nltk (#50)

Add alembic directory to exclusion list

Add .tsv files to exclusion list

Add "migrations" folder to exclusion list

Fix Marqo namespace issue. (#45)

fix merge conflict

sage env

Pass access_token in the constructor of DataManagers.

move punkt download to vector store where it's actually used

Update vector_store.py

Take "namespace" out of the vector store constructors.

download nltk if not downloaded

Retrieval benchmark (#39)

Add YAML configurations (#38)

Split --index-name flag into --pinecone-index-name and --index-namespace.

Option to index repo at specific commit.

Add Cohere, NVIDIA, Jina & HuggingFace rerankers (#37)

Add option for Pinecone + BM25 hybrid retrieval. (#36)

Now that we have a reranker, retrieve 25 items from the vector DB.

Streaming responses.

Add Hugging Face cross-encoders

Add Cohere reranker

Simplify name to "Sage" (#35)