HOW TO BUILD A Multi-document RAG Chatbot FOR BEGINNERS | Deploy your AI streamlit App for free

#streamlit #generativeai #langchain #chatbot #openai #nlp
How to build a Multi-doc chatbot using Retrieval Augmented Generation (RAG), LangChain, openai & Streamlit - Full tutorial end-end.

With this video, you will be able to build a document Q&A chatbot using the RAG framework. This video also provides langchain tutorial.

You will understand.
- langchain chatbot
- how llms work langchain
- nlp projects
- openai APIs
- how to get an openai api key for free
- what is Retrieval Augmented Generation (RAG)
- what are openai embeddings
- what are vector databases
- deploy streamlit app
- Streamlit python tutorial

Chapters :

00:00 Introduction to the app
01:00 How the App Works
02:00 Understanding Langhain
03:00 Langchain functions and classes
07:20 Code overview
12:00 Deploying the App on Streamlit Community Cloud


#langchain #streamlit #python #generativeai #openaiapi #ai #ragframework

👨 WHO AM I -

I'm Sri Laxmi an AI product Manager who lives in San Francisco, CA. On this channel, we will learn how to build generative AI applications and use AI tools that can help us launch the projects that inspire us and, consequentially, lead the lives we've always dreamed about.

Sneak peek of upcoming projects:

- *Text Summarization:* Harness the power of Langchain and OpenAI to distill essential information from extensive texts, making comprehension faster and more efficient.
- *Llama 2 Chat:* Dive into conversational AI with a chatbot built on Meta's open-source Llama 2 LLM, designed for dynamic and engaging interactions.
- *Social Media Toolkit:* Elevate your online presence with a Hashtag and Caption Generator, utilizing the Cohere API for creative and impactful social media content.

If that sounds interesting, consider subscribing! See you in the next video 😀

Linkedin -

[spotnuru]

thank you for this but can you do with all opensource technologies and share a video, dont want to use anything which charges us, thanks in advance.

[uwegenosdude]

Hello ! Thanks for the video. I wanted to try it out but first I got an error, cause llms, embeddings and vectorstores are now part of langchain_community, instead of langchain.
After changing this, the streamlit started. But after uploading a PDF (in your video you show txt files) I get an decode error.
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 158: invalid start byte
Did you ever try it with PDF?
OK, I had to run "pip install --upgrade langchain".
And I changed thre streamlit_app from pdf to txt.
Now I can upload docs und ask a question. But even for small documents I get an length error like:
This model's maximum context length is 4097 tokens, however you requested 8278 tokens
Can I somewhere allow more tokens?
Or can I choose another model which has a higher context length?
OK, with gpt-4... I fixed the problem with the context length.
But now I get another error:
You tried to access openai.ChatCompletion, but this is no longer supported in openai>=1.0.0
You can run `openai migrate` .... or you can pin your installation to the old version, e.g. `pip install openai==0.28`
Are you using this old version?
OK, update: I downgraded to openai 0.28.
Then after eliminating the openai_api_key=openai_api_key param in I could really ask questions about my uploaded txt file.
Could you perhaps create an updated version using openai 1.x.y ?
Thanks a lot in advance. Great work !
Just another update: I'm fascinated about the high quality of the answers. Much better than any custom GPT with uploaded files.
Could I also upload txt-files that contain JSON? Cause I crawled our website with gpt-crawler.

[PrabhatKumar-nrys]

Hi have we done video on extracting tables and images from source document and display as output in rag system.? Please help