Building Contextual ChatBot using LLMs, Vector Databases & Python
summary
Demo shows a runnable Python pipeline that turns PDFs into chunks, creates embeddings with text-embedding-ada-002, stores them in ChromaDB, and queries via a LangChain retriever feeding gpt-3.5-turbo with memory. Code includes splitter (chunk size ~2550, overlap ~250), Streamlit front end, and examples of temperature=0 for factual answers. Addresses token limits, RAG workflow, and low API cost estimates so viewers can reproduce the contextual chatbot end-to-end.
AI-written summary · about
chapters (15)
- 0:00Building contextual chatbots
- 1:01Generative AI with LLMs
- 4:58How LLMs work
- 9:16LLMs in business and coding
- 12:46Designing chatbot with LLMs
- 17:07Prompt engineering
- 21:22Demo: Basic prompting with OpenAI
- 33:53LLM challenges and limits
- 37:29LangChain for documents
- 41:19Vector databases and embeddings
- 45:55Chatbot with LangChain & ChromaDB
- 1:00:27Memory and retrieval in chatbots
- 1:12:06Practical chatbot limits
- 1:19:31Advantages and disadvantages of LLMs
- 1:24:46LLM future and next steps
related talks
01h 12mBuild Your Own AI Chatbot with OpenAI & LangChain
21mChatGPT, LLMs, and LangChains - A Beginner's Guide
25mLets Build A Node.js ChatGPT-Like Bot With AI APIs
47mEnd-to-End Chatbot Web App with Your Own Data
58mDemystifying GenAI: Build a ChatGPT App with Vector Store
31mMachine Learning-Enabled Chatbots
15mBuilding Your First Chatbot with LangChain and IBM's Granite LLM Model
28mWhat the RAG? Demystifying Retrieval Augemented Generation