Skip to content

folder_structure.md

Latest commit

 

History

History
113 lines (91 loc) · 3.73 KB

folder_structure.md

File metadata and controls

113 lines (91 loc) · 3.73 KB
Raw

Folder Structure

This document describes the folder structure of the LLM Backend Framework.

Overview

The project is organized into several key directories:

├── api/                  # API layer with FastAPI
│   ├── v1/               # Versioned APIs
│   ├── middleware/       # API middleware components
│   ├── app.py            # API application configuration
│   └── __init__.py       # API initialization
├── src/                  # Source code
│   ├── chat_engine.py    # Chat engine implementation
│   ├── config_injector.py # Dependency injection setup
│   ├── base/             # Core framework pieces
│   │   ├── brains/       # Brain interface and variants
│   │   └── components/   # Modular building blocks
│   │       ├── llms/     # LLM client implementations
│   │       ├── memories/ # Conversation memory backends
│   │       ├── embeddings/ # Embedding generators
│   │       ├── tools/    # Tool implementations
│   │       └── vector_databases/ # Vector DB integrations
│   ├── experts/          # Domain-specific chat experts
│   │   ├── qna/          # Q&A expert implementation
│   │   └── rag_bot/      # Retrieval‑augmented generation expert
│   └── common/           # Shared utilities
├── tests/               # Test directory
├── docs/                # Documentation
├── app.py               # FastAPI application entry point
├── cli.py               # Command-line interface
├── Dockerfile           # Docker configuration
├── Makefile            # Build and development commands
├── pyproject.toml      # Python project configuration
└── requirements.txt    # Project dependencies

Key Components

API Layer (api/)

The API layer handles HTTP requests and responses using FastAPI:

  • api/v1/ - Versioned API endpoints
  • api/middleware/ - Custom middleware components
  • api/app.py - FastAPI application configuration
  • api/__init__.py - API initialization

Source Code (src/)

Chat Engine

  • src/chat_engine.py - Chat engine implementation for message processing
  • src/config_injector.py - Dependency injection configuration

Base Layer (src/base/)

  • src/base/brains/ - Brain interfaces and variants
  • src/base/components/ - Core components
    • llms/ - LLM client implementations
    • memories/ - Conversation memory implementations
    • embeddings/ - Embedding generators
    • tools/ - Tool implementations
    • vector_databases/ - Vector database integrations
    • README.md - Components documentation

Experts (src/experts/)

  • Domain-specific experts built on top of the base layer
  • qna/ - Question answering expert
  • rag_bot/ - Retrieval‑augmented generation expert

Common Utilities (src/common/)

  • Shared utilities and models
  • Configuration management
  • Logging setup

Testing (tests/)

  • Unit tests
  • Integration tests
  • Test fixtures and utilities

Documentation (docs/)

  • api.md - API documentation
  • folder_structure.md - This document

Architecture Overview

The application follows a layered architecture:

  1. API Layer

    • Handles HTTP requests/responses
    • Input validation
    • Error handling
    • Middleware processing

  2. Base Layer

    • Brain abstractions and component factories
    • Chat engine integration

  3. Experts

    • Domain-specific logic built on the base layer

  4. Common Layer

    • Shared utilities
    • Configuration management
    • Logging

This structure promotes:

  • Clear separation of concerns
  • Modularity and extensibility
  • Easy testing and maintenance
  • Consistent error handling
  • Comprehensive logging