How mahmoudmabrok/chatwithcode works
Main Components Identification
a) User Interface Layer:
• Streamlit Web Interface (from app.py)
b) Core Processing Layer:
• LLM Integration (Ollama with deepseek-coder-v2 model)
• Vector Store Management (FAISS)
• Embedding System (HuggingFaceEmbeddings)
• Document Processing (LangChain)
c) Storage Layer:
• FAISS Vector Store (vector_store_index directory)
• Local Git Repository (external input)
[User Interface Layer]
• ↓
[Core Processing Layer]
• ↓
[Storage Layer]
Detailed Component Representation
• Rectangles: Main components (UI, Processing, Storage)
• Hexagons: External services (Ollama, Local Git Repo)
• Rounded rectangles: Processing components (Embeddings, Document Processor)
Data Flow Representation
Two main flows:
a) Knowledge Base Creation Flow:
Local Git Repo → Document Processor → Embeddings → FAISS Storage
b) Query Flow:
User Input → LLM → Vector Store Query → Response Generation → UI
Component Details
For each component, include:
• Component name
• Primary function
• Key technologies used (e.g., "Streamlit UI", "FAISS Vector Store")