# LangChain Learning Demos
Personal learning repository for LangChain examples and demonstrations.

## Learning Resources
- [Building AI Apps with LangChain](https://youtu.be/yF9kGESAi3M?feature=shared)
- [LangChain Tutorial](https://youtu.be/i-oHvHejdsc?feature=shared)

## Tech Used
- LangChain - Python framework for building LLM applications
- Vector Databases - For storing and retrieving text embeddings
    - PineCone 
    - AstraDB (single demo)
- Text Splitters:
    - CharacterTextSplitter
    - RecursiveCharacterTextSplitter
    - TokenTextSplitter
- Embeddings:
    - OpenAI Embeddings
    - HuggingFace Embeddings
- Retrievers:
    - Similarity Search
    - MMR (Maximal Marginal Relevance)
    - Time-weighted Vector Store
- Loading Tools:
    - TextLoader - For processing text files
    - WebBaseLoader - For loading web content
    - FireCrawl - For web crawling
- Large Language Models:
    - OpenAI
        - GPT-4
        - GPT-3.5-turbo
    - Anthropic
        - Claude-3-Sonnet
    - Google
        - Gemini 2.0
    - NVIDIA
        - Mixtral 8x22B
    - Groq
        - Mixtral 8x7B

## Overall Process Flow
```mermaid
graph LR;
    A[Read Text] --> B[Split into Chunks];
    B --> C[Store in Vector DB];
    C --> D[Retrieve from Vector DB];
    D --> E[Send to GAI Model];
```

A. Read Text
    - Text files (TextLoader)
    - Web content (WebBaseLoader, FireCrawl)

B. Split into Chunks
    - The chunking strategy plays an important role. #3 has different examples.

D. Retrieve from Vector DB
    - There are different retrieval strategies. #5 has different examples.