rag.md

RAG Architecture

Challenges

1. Context embeddings creation

   • how are embeddings created? which strategy was used?

2. Query + context

   • Ensure query + context reflect what's intended
   • provides an acceptable answer

3. Include the generated answer in the correct place in the project

RAG General view

flowchart
    
    subgraph Prompt
        A[/Query/] --> B[Embedding model<br> <strong>text-embedding-ada-002</strong>]
        B --> C[Embeddings]
    end
    subgraph Context
        E[/Data source/] --> F[Embedding model<br> <strong>text-embedding-ada-002</strong>]
        F --> G[Embeddings]
    end
    subgraph Vector Database
        C --> D[(<strong>PostgreSQL PGVector</strong>)]
        G --> D
    end
    subgraph LLM
        A --> H[/Query + Context/]
        D --> H
        H --> I[OpenAPI o1]
        I --> J[/Response/]
    end

Loading

Integrations

flowchart
    subgraph CodeMonkey
        API
    end

    subgraph Local
        C[Repository]
    end

    subgraph Github
        D[Repository]
        Issues
        App
    end

    Github <--> CodeMonkey
    Local <--> CodeMonkey

Loading

Embeddings and LLM

Embeddings	LLM
text-embedding-ada-002	OpenAPI o1
text-embedding-ada-002	Llama 3.2 (local)