RAG Based Chatbot With Memory(Chat History) | Creating History Aware Retriever | Langchain

[@thesacredcode233]

Thank you very very much champ🏆🔥

[@sunnysavita10]

Thankyou 😊

[@AmmaraIlyas-hz3wj]

Thats very great session, I have been searching for the best playlist for quite some time, and I guess this playlist will prove to be very helpful💓

[@sunnysavita10]

Glad you liked it

[@shawtylikesfood]

Thank you so much for putting in so much effort into these please keep them coming💗💗

[@sunnysavita10]

sure why not thankyou

[@Ishaheennabi]

Great session keep going ❤❤❤

[@PhotoshoppersStop]

Thank you so much Sir, It was a long awaited one for me 😇😇

[@sunnysavita10]

Most welcome 😊

[@Sandeepsavita10]

Great session Sunny 🎉🎉🎉

[@jamilahmarramduarte1575]

BEST EXPLAINATION!

[@RaviRanjan-vg8pe]

Thank you so much sir
I am looking for this only 🙏🙏🙏

[@sunnysavita10]

Most welcome

[@MannDakuwa-p9s]

This is done with few shot learning prompt right ?

[@sunnysavita10]

i did not get you

[@amansingh-ih2by]

How can we do the same thing with agent executor class? And how to check whether the context of the rag is passed or not

[@sunnysavita10]

soon i will create a video ontop of agentic RAG with memory then you will get a idea

[@arshahin9788]

As a beginner how to start langchain form your channel?

[@sunnysavita10]

start with langchain crash course

[@Rr-lh2ru]

Sir please make a Deep learning playlist

[@sunnysavita10]

soon will uploaded

[@anandukc4709]

I have a doubt. If the question doesn't needs any chat history the same process will repeat right? So its actually wasting tokens? Is there any ways to solve this??. And also how to manage session in real-life like in production there may be several users right so how to handle that situation? Pls reply

[@sougaaat]

What I found problematic here is the usage of continuous chat history, in such a way as you said, we're wasting tokens, there was `ConversationSummaryMemory` in the LangChain framework to provide a summary of the chat history but sadly LangChain has discontinued support for that and advises users to migrate to LangGraph persistence which is a little complicated. Secondly, I don't think there's any way to decide beforehand if a user query requires the chat history context, so, that's why it's more robust to at least provide a summary of the chat history as an additional context for more accurate responses.

[@anandukc4709]

@@sougaaat true langraph documentation itself is very complicated. But one thing that i done is message trimming to reduce token usage.

[@sougaaat]

@@anandukc4709 Indeed. It's really irritating how fast they introduce breaking changes.

[@Abhishekkumar-jl2dr]

Which LLM should I use sir to run locally ?

[@sunnysavita10]

watch my ollama video it is having plenty of options

[@rakeshkumarrout2629]

this is great,but what is session_id still a mystery for me.can you shed some light? some answer and link.

[@tomtom-it4co]

A session ID in generative AI typically refers to a unique identifier assigned to a user session. It tracks the interaction context between the user and the AI model, allowing the model to maintain continuity and reference previous inputs, ensuring coherent responses throughout the session.

[@rakeshkumarrout2629]

@@tomtom-it4co thanks for sharing.bdw how can we track that?

[@rakeshkumarrout2629]

@@tomtom-it4co where can we get that from?

[@sunnysavita10]

it could be any random value

[@rakeshkumarrout2629]

@@sunnysavita10 okh got it ,bdw how to track for each user when there is 1000s of user??