DistanceStrategy used for faiss

[josh-zhang]

In faiss_cache.py module, when init faiss instance, the current code is using METRIC_INNER_PRODUCT as distance_strategy, shouldn't this be 'MAX_INNER_PRODUCT'?

since there is no METRIC_INNER_PRODUCT in langchain_community.vectorstores.utils.DistanceStrategy, only:

  EUCLIDEAN_DISTANCE = "EUCLIDEAN_DISTANCE"
  MAX_INNER_PRODUCT = "MAX_INNER_PRODUCT"
  DOT_PRODUCT = "DOT_PRODUCT"
  JACCARD = "JACCARD"
  COSINE = "COSINE"

besides, in langchain_community.vectorstores.faiss, during index initialisation, we can see the logic as below:

    if distance_strategy == DistanceStrategy.MAX_INNER_PRODUCT:
        index = faiss.IndexFlatIP(len(embeddings[0]))
    else:
        # Default to L2, currently other metric types not initialized.
        index = faiss.IndexFlatL2(len(embeddings[0]))

[dosubot[bot]]

🤖

It's great to see your interest in contributing to Langchain-Chatchat! Your observation about the distance strategy in the faiss_cache.py module is valuable. If you believe that using 'MAX_INNER_PRODUCT' instead of 'METRIC_INNER_PRODUCT' would be more appropriate, I encourage you to go ahead and create a pull request with your proposed change. Your contribution will help improve the accuracy and efficiency of Langchain-Chatchat. Thank you for considering to contribute!

---

This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

If you want to continue the conversation, start your reply with @dosu-bot.