You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
langchain/libs
Kenneth Choe 62efe1ffb9
support add_embeddings for elasticsearch (#11002)
- **Description:** Provide a way to use different text for embedding.
- For example, if you are ingesting stack-overflow Q&As for RAG, you
would want to embed the questions and return the answer(s) for the hits.
With this change, the consumer of langchain can implement that easily.
- I noticed the similar function is added on faiss.py with #1912 which
was for performance reason, but I see the same function can be used to
achieve what I thought. So instead of changing Document class to have
embedding_content, I mimicked the implementation of faiss.py.
- The test should provide some guidance on how to use it. It would be
more intuitive if I just pass texts and embedding_texts as separate
arguments, but I chose to use `zip`-ed object for the consistency with
faiss.py implementation.
      - I plan to make similar pull request for OpenSearch.
  - **Issue:** N/A
  - **Dependencies:** None other than the existing ones.

Co-authored-by: Bagatur <baskaryan@gmail.com>
11 months ago
..
experimental bump 317 (#11986) 11 months ago
langchain support add_embeddings for elasticsearch (#11002) 11 months ago