Pinecone + Cohere Reranker

Depends on the use case here? Did you look into trying it out on some of the queries on your documents and see if it’s helping you get better relevant nodes?

https://colab.research.google.com/drive/1DNL5-rxa8oYtgJOHb00NK6nnEQ74ObTe?usp=sharing

have checked above and this notebook https://github.com/jerryjliu/llama_index/blob/v0.6.1/examples/node_postprocessor/CohereRerank.ipynb
do we have any documentation that shows uplift in quality by using CohearAI along with pinecone? do you suggest that we should test ourself to get a better analysis of our usecase?

Yes, it's better to test on your use-case and documents. As it entirely depends on the data you are using.

Can Anyone help me, please? I am working on a project, where I have too many movie's description, including movie name, type of movie(romantic, horror etc) description etc. I have all this data in form of text. I have too many text files each movies.

Now, the thing I want to perform here is:

1. Split this data using textSplitter.

1. Store in pineconeStore with openai gpt-3.5 embedding.

1. Now, for example when user ask for: "show me list of romantic movies" then, provide him list of romantic movies.

Issue: Because of there is no relationship between the split documents, and having movie type and movie name in different documents, I get the result that there is a romantic movie, "but as the name is in different document, I am not able to get the name of the movie."

Can you please help me out regarding this?

I would split each movie into its own document before indexing, then set the top k to something like 10