Next-Level ReRanking with FlashRank: A Speedy Solution for Advanced RAG

[@bhosh]

Man, this guy knows everything! Fantastic! - kudos from San Francisco

[@akki_the_tecki]

Hi sonu, Make a complete Video from uploading PDF Docs to getting the valid responses too,. We want the code too 🎊🎊🎊🎊✨✨
Waiting for this

[@SonGoku-pc7jl]

thanks for all your words!!! :D

[@AIAnytime]

My pleasure!!

[@MarceloLimaXP]

I ran the app here, even using Large the response time is absurdly fast =)

[@AIAnytime]

Yes it is impressive

[@abhirampemmaraju6339]

What about accuracy?

[@noob.gamers697]

Hi,
Thankyou for this great lecture.
When I was using reranking for my use case so that I can give relevant information . I was successful in that.
but I am using that result using ConversationalRetrievalChain as replacing my ranking result with the retriever . I am getting an error
"ValidationError: 1 validation error for ConversationalRetrievalChain
retriever
value is not a valid dict (type=type_error.dict)" can you tell me how can I resove this

[@kumarmangalam9170]

Does flashrank perform better than langchain component

[@dinushdealwis9687]

We can use this reranker post-retrieval correct?

[@seththunder2077]

Can we use this with 3.5 GPT and langchain to improve my rag? Tbh im struggling with which advanced rag to use as I have 22 pdf files. Which one do you recommend

[@wilfredomartel7781]

🎉 how to fine tuning multiBert reranker? Please any help

[@matten_zero]

Is RAG dead? I'm hearing that the actual WAY is LoRA and imbedding knowledge into the weights via fine tuning layers.
Or maybe both?

[@gsayesh]

Dear Sir, Please Upload (Get Started with Qdrant Vector Database: Build your First RAG (Part 2)) I'm waiting for Very Advanced RAG Tutorial within one video. Please upload sir very advanced level rag with RHLF tutorial.

[@akki_the_tecki]

hey bro, could you please help me with what LLM and Embedding Model should we use to have accurate response (I'm having 500 PDFs)

[@MachineLearningZuu]

Bro it's very hard to scale for that much amount. but you can try finetuning embeddings. @@akki_the_tecki

[@wealth_developer_researcher]

Thanks you so much for such a nice video

[@freeuser7701]

This tells me nothing, Why not use BGE model instead of reranker, Re ranker needs larger model than current Context embedding model.

[@abhirampemmaraju6339]

Bge models is having latency. They are taking too much time although accuracy is almost equal to cohere